Advanced Micro Devices (AMD) stock closed the May 22 trading session 3.99% up at $467.51, hitting its all-time high.
The stock has rallied and made impressive year-to-date gains of 118.3% at the time of writing, Saturday morning, May 23.
The rally has been boosted by hyperscalers increasing their capital expenditure plans and Intel’s Q1 earnings, which have boosted confidence in the semiconductor sector.
On May 21, AMD confirmed investments of more than $10 billion across Taiwan’s semiconductor ecosystem to improve leading-edge silicon, packaging, and manufacturing technologies.
These manufacturing innovations are necessary to support the deployment of the AMD Helios rack-scale platform in the second half of 2026.
AMD has also confirmed something equally important: it has beaten other CPU manufacturers in the race for the best manufacturing process.
AMD is ramping production of its next-generation EPYC Processor “Venice” on an advanced 2nm process.Shutterstock
AMD ramps production of its next-generation EPYC Processor
AMD is ramping production of the next-generation AMD EPYC processor in Taiwan, manufactured on Taiwan Semiconductor Manufacturing Company’s (TSMC) advanced 2nm process (node) technology.
The latest incarnation of EPYC CPU is codenamed “Venice,” and is a key part of AMD Helios-based systems, along with AMD Instinct MI450 GPUs. AMD also plans to ramp production of these chips at TSMC’s Arizona fabrication facility in the future.
The execution of the AMD Helios ramp-up is one, if not the main, downside risk for AMD. The platform is designed to be a direct competitor to Nvidia’s Vera Rubin NVL72 racks. This is why this ramp of EPYC is of critical importance for AMD this year.
According to AMD, “Venice” is the first high-performance computing product in the industry to enter production on TSMC’s 2nm node.
The company aims to extend TSMC’s 2nm process technology across its data center CPU roadmap, with future versions of EPYC, codenamed “Verano,” optimized for performance-per-dollar-per-watt.
TSMC currently has the best node when yields and capacity are taken into account
Manufacturers typically do not divulge their yields.
Heqing Electronics said in January 2026 that initial TSMC 2nm yield rates are reportedly reaching around 70%, with some sources indicating yields surpassing 90% for certain memory products.
The ramp of EPYC signals that reports TSMC had very good initial yields have been correct. This makes TSMC’s 2nm the best node when yield, and capacity are taken into account.
The node is important because its size indicates device generation, with smaller features and higher transistor density, resulting in improved performance and efficiency.
It is important to note that while the node matters a lot, the CPU architecture itself still matters more.
Nvidia still on the 3nm node, and Intel has constrained capacity
“Vera CPU opens a brand-new $200 billion [total addressable market] for Nvidia, a market we have never addressed before. Every major hyperscale and system maker is partnering with us to get it deployed. We have visibility to nearly $20 billion in total CPU revenue this year, setting us up to become the [world ’s] leading CPU supplier.”
As Kress explained, the specialized chips are Nvidia’s entry into the CPU data center market and are set to compete with Intel’s Xeon and AMD EPYC CPUs.
Nvidia has never officially stated which node it is manufacturing Vera on, but if it were the top node, the company wouldn’t have avoided mentioning it. According to both Tom’s Hardware and Semianalysis, Vera Rubin is manufactured on a 3nm-class TSMC process.
On paper, Intel’s (INTC) 18A is the best node. However, in practice, the company is capacity-constrained and is still working to improve yields.
During the Q3 2025 earnings call, Intel CFO David Zinsner talked about the yields:
“I would say the yields are adequate to address the supply, but they are not where we need them to be in order to drive the appropriate level of margins,” Zinsner said. “By the end of next year, we’ll probably be in that space. Certainly, the year after that, I think they’ll be in what would be kind of an industry-acceptable level on the yields.”
More Tech Stocks:
The situation seems to have improved, as at the Q1 2026 earnings call, Intel CEO Lip-Bu Tan said: “We have made steady progress with Intel 4 and Intel 3, and 18A yields are now running ahead of the internal projections, representing a meaningful inflection in our execution and our factory finished goods output.”
And while the company might be ahead of its 2027 internal projection for good yields, it likely still has a few more months before it reaches that level, and its capacity still needs to be improved.
According to Bank of America, as Intel management has said numerous times, additional wafer customers would require additional capex, and that meaningful volume might only start in 2028 and beyond.
Analysts raised their estimates for the total addressable market for data center server CPUs to $110 billion from $80 billion.
Arya reiterated a buy rating for AMD stock and raised the target price to $500 from $450, based on a 42x multiple of his 2027 non-GAAP EPS estimate.
Goldman Sachs analyst James Schneiderupdated his opinion on AMD stock following the earnings report. He upgraded AMD stock to buy from neutral and raised the price target to $450 from $240.
According to TipRanks, 28 of the 36 analysts covering AMD stock rate it a buy. Eight give a hold rating. The average price target is $461.38.
Downside risks for AMD:
Execution on the first rack-scale product (MI400 Series)
